Many serious health problems, particularly those associated with aging, can be improved by regular exercise, yet many Americans struggle to establish a consistent exercise habit. One-third of Americans today do not participate in any leisure-time physical activity. As a result, health officials and policy makers have a keen interes in interventions that can promote exercise. This proposal takes a close look at commitment contracts - an innovative, but not-yet-widely-studied incentive program. Commitment contracts offer individuals the opportunity to set a goal and put their money at stake to reach that goal. If the person meets their goal, they get their money back, but if they do not succeed, they lose the money. Behavioral economic theory suggests that these contracts are useful for overcoming issues of self-control. Additionally, this theory predicts that the desire for commitment is strongr the further in the future the commitment begins. We test these assertions by addressing the following specific aims: 1) to conduct a survey to better understand the demand for commitment contracts and their effectiveness, 2) measure the fraction of people interested in creating commitment contracts and test whether demand for them varies with contract starting date, and 3) estimate the effect of commitment contracts on exercise behaviors and the relative effectiveness of different commitment contracts. Under a unique partnership with a fitness gym consisting of 3,100 members, we propose to conduct a randomized real-world field experiment to study the effectiveness of these contracts. The population of gym members is particularly relevant for studying commitment because the purchase of a membership demonstrates a clear interest in exercising more, yet our preliminary data show that many fail to follow through with this intention. We plan to study two types of contracts: a standard commitment contract with an immediate start date and a new type of commitment contract, a delayed commitment contract with a start date one month in the future. Our intervention combines gym-use data with unique survey measures, including information on self-efficacy, self-control, and time budgeting. We use these measures to better understand how individual characteristics predict commitment contract take-up. Our findings then can be used to target these contracts to individuals who may most benefit from them. Although our focus is on improving exercise behaviors, since problems of self-control are so pervasive in other dimensions of health behavior, commitment contracts may find wider application. Relative to other types of incentive programs, commitment contracts are appealing because they are low-cost, as they do not rely on external funds, and promote individual responsibility for setting and achieving health goals rather than relying on externally imposed health targets. Thus, they may be implemented in settings where few resources are available for changing health behavior.